A Common Pharmacophore for Cytotoxic Natural Products That Stabilize Microtubules

Proc. Natl. Acad. Sci. USA Vol. 96, pp. 4256–4261, April 1999 Chemistry

A common pharmacophore for cytotoxic natural products that stabilize microtubules

IWAO OJIMA*†,SUBRATA CHAKRAVARTY*, TADASHI INOUE*, SONGNIAN LIN*, LIFENG HE‡,SUSAN BAND HORWITZ‡, SCOTT D. KUDUK§, AND SAMUEL J. DANISHEFSKY§

*Department of Chemistry, State University of New York at Stony Brook, Stony Brook, NY 11794-3400; ‡Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY 10461; and §Laboratory for Bioorganic Chemistry, Sloan Kettering Institute for Cancer Research, 1275 York Avenue, New York, NY 10021

Contributed by Samuel J. Danishefsky, February 22, 1999

ABSTRACT Taxol (paclitaxel), a complex diterpene ob- hancement (NOE) spectroscopy experiment at 4°C in DMSO- ͞ ϭ 2 tained from the Pacific yew, Taxus brevifolia, is arguably the most d6 D2O (3:1; D2O H2O) with solvent peak suppression important new drug in cancer chemotherapy. The mechanism of (Fig. 2), carried out on a Varian INOVA 500-MHz spectrom- cytotoxic action for paclitaxel—i.e., the stabilization of micro- eter. Simulated annealing of several starting structures of 5 (10 tubules leading to mitotic arrest—is now shared by four recently cycles each of the following sequence per conformer: equili- identified natural products, eleutherobin, epothilones A and B, bration at 1000 K for 2 ps, followed by exponential cooling and discodermolide. Their ability to competitively inhibit from 1000 K to 300 K over 10 ps) including the key NOE-based [3H]paclitaxel binding to microtubules strongly suggests the constraints for the C-2 and C-3Ј side chains was carried out existence of a common binding site. Recently, we have developed (SYBYL 6.4), and the lowest-energy conformer consistent with nonaromatic analogues of paclitaxel that maintain high cytotox- these NOEs was chosen as the template conformation for the icity and tubulin binding (e.g., nonataxel). We now propose a superimposition study. common pharmacophore that unites paclitaxel, nonataxel, the Several low-energy conformations of epothilone B (2b), epothilones, eleutherobin, and discodermolide, and rationalizes eleutherobin (3), and discodermolide (4) obtained from the the extensive structure–activity relationship data pertinent to high-temperature molecular dynamics study (SYBYL 6.4, Tripos these compounds. Insights from the common pharmacophore force field, charges calculated by the Gasteiger–Hu¨ckel have enabled the development of a hybrid construct with dem- method) were compared with the template conformation of 5. onstrated cytotoxic and tubulin-binding activity. It should be noted that the purpose of this study was not to obtain a statistical distribution of the various possible overlays, Taxol (paclitaxel, 1; see Fig. 1) is a powerful resource in cancer but to arrive at the best fit that generally satisfied the vast chemotherapy (1–7). It was approved by the Food and Drug existing SAR and structural data for these compounds.** Administration in the United States for the treatment of ad- Numerous overlay modes were generated, starting from these vanced ovarian cancer (1992) and metastatic breast cancer (1994). Clinical studies on the treatment of other cancers with Abbreviations: SAR, structure–activity relationship(s); NOE, nuclear paclitaxel and closely related synthetic analogues, as well as Overhauser enhancement. combination protocols, are being pursued (1–7). †To whom reprint requests should be addressed. e-mail: iojima@ Recently, four natural products, structurally dissimilar to pac- notes.cc.sunysb.edu. ¶We note that, recently, the structure of the ␣␤-tubulin dimer incor- litaxel, have been found to share its mechanism of action (8, 9). ϭ ϭ porating a paclitaxel molecule by electron crystallography of zinc- Epothilones A (2a: R H) and B (2b: R Me) (10, 11), induced tubulin sheets was reported with 3.7-Å resolution [see eleutherobin (3) (12), and discodermolide (4) (13) (Fig. 1) show Nogales et al. (29)]. A proposal for the tubulin-bound structure of activities comparable to those of paclitaxel in various assays paclitaxel, however, was not conclusive because of insufficient reso- (11–15). These compounds competitively inhibit [3H]paclitaxel lution. If a highly reliable refined crystal structure of tubulin-bound paclitaxel becomes available, docking experiments of these microtu- binding to microtubules (11, 16, 17), strongly suggesting the bule-stabilizing drugs with the protein binding pocket can be per- ¶ existence of a common binding site. Clearly, the identification of formed to define their common pharmacophore. In the absence of a three-dimensional pharmacophore common to these structur- such reliable information, our approach is directed toward identifying ally diverse agents could guide the selection of the next generation common structural features among these drug molecules that corre- late for their tubulin-binding ability. of paclitaxel-like cytotoxic agents. Herein we propose a com- ʈ mon pharmacophore for these agents based on their structural Prior preliminary studies on the common pharmacophores for nonataxel, paclitaxel, epothilones, and discodermolide have been pre- correlation to the NMR-defined conformation of the nonaro- ʈ sented at (i) the 214th American Chemical Society National Meeting, matic paclitaxel mimic nonataxel (5). The pharmacophore Las Vegas, NV, Sept. 7–11, 1997, by S. Chakravarty and I. Ojima, succeeds in explaining the substantial structure–activity rela- abstr. MEDI 75; and (ii) the 215th American Chemical Society tionships (SAR) of these agents and is corroborated by the National Meeting, Dallas, TX, March 29-April 2, 1998, by S. F. synthesis of a hybrid construct with demonstrated cytotoxic Victory, G. L. Grunewald, and G. I. Georg, abstr. MEDI 187. **The proposed bioactive conformation of epothilone B was obtained and tubulin-binding ability. through the template forcing protocol (DISCOVER 95.0), using the whole southern region of the nonataxel structure derived from the NOE- MATERIALS AND METHODS constrained molecular dynamics as the template. This structure is Conformational Analysis of Nonataxel and Molecular Mod- among the lowest-energy conformers generated through this opera- eling. Nonataxel (5) was subjected to conformational analysis tion, albeit not the global energy minimum structure. It is worth mentioning that the hypothetical bioactive conformation of epothilone using distance constraints based on the cross-peak intensity B thus obtained is found to be very close to its x-ray crystallographic obtained from the two-dimensional nuclear Overhauser en- structure [see Ho¨fle et al. (30)], and only slight deviation is observed at the orientation of the lactone carbonyl. For eleutherobin, the The publication costs of this article were defrayed in part by page charge molecular dynamics calculations were carried out without any constraint because of its highly rigid structure except for its side chains. The payment. This article must therefore be hereby marked ‘‘advertisement’’ in positions of the two side chains at C-3 and C-8 were defined by using accordance with 18 U.S.C. §1734 solely to indicate this fact. the t-Boc group at C-3Ј-N and the acetyl group at C-10, respectively, as PNAS is available online at www.pnas.org. key positions in the template forcing operations.

4256 Downloaded by guest on September 27, 2021 Chemistry: Ojima et al. Proc. Natl. Acad. Sci. USA 96 (1999) 4257

FIG. 1. Structures of paclitaxel (1), epothilones A and B (2a and 2b), eleutherobin (3), discodermolide (4), and nonataxel (5). Labeled boxed regions are areas of common overlap as represented in Fig. 4.

conformers, that were optimized with template forcing studies proposed by Winkler and Axelsen (18) for paclitaxel and [DISCOVER 95.0, consistent valence force field (CVFF)].†† To epothilone B (see below). refine the structures resulting from the template forcing Synthesis of Hybrid Construct 6. Fig. 3 outlines the synthesis procedure, the models were subjected to restrained and un- of hybrid construct 6 starting from 7,10,13-tri-TES-1,2- restrained molecular dynamics (SYBYL 6.4, Tripos force field carbonate baccatin 7. Nucleophilic ring-opening of the car- using Gasteiger–Hu¨ckel charges), typically for 100 ps for each bonate moiety of 7 by using the Grignard reagent from conformer at 1000 K with sampling at every 1 ps, followed by 2-allyloxyphenyl bromide yields baccatin 8 in 79% yield. dynamics at 300 K for 5 ps for each sampled conformer. Each Deprotection of the TES groups with HF͞pyridine, followed of these conformers was then minimized and the resultant by reprotection at the 7-position with TES and acetylation at low-energy conformers within the range of 10 kcal͞mol were the 10-position, gave baccatin 8 in 56% yield over three steps. reevaluated for overlay with the template nonataxel confor- Coupling of baccatin 9 with ␤-lactam 10, obtained in good mation. Similar analysis was performed for the overlay mode yields and high enantiomeric excess by the previously published procedure (6, 19), provided coupling product 11 in 64% yield. The ring-closing metathesis of 11 with Ru–benzylidene †† While we used two different force fields—i.e., Tripos (SYBYL 6.4) and complex in CH2Cl2 (1 mM) proceeded smoothly to afford the CVFF (DISCOVER 95.0)—all energy optimizations were performed protected 18-membered macrocycle as the pure E isomer, with the Tripos force field. CVFF was used only for template forcing ͞ operations. which afforded hybrid construct 6 on deprotection with HF

FIG.2. (Right) Two-dimensional NOE spectrum in DMSO͞D2O (3:1) for nonataxel (5) at 4°C (500-MHz Varian INOVA.) In table on Left, NOE intensities and corresponding distance constraints: s ϭ strong (1.8–2.5 Å); ms ϭ medium-strong (1.8–3.0 Å); m ϭ medium (1.8–3.5 Å); mw ϭ medium-weak (1.8–4.0 Å); w ϭ weak (1.8–5.0 Å); —- ϭ no observed NOE; ND ϭ not quantifiable. Downloaded by guest on September 27, 2021 4258 Chemistry: Ojima et al. Proc. Natl. Acad. Sci. USA 96 (1999)

FIG. 3. Synthetic scheme for the hybrid construct 6 (SB-TE-1120). TES, triethylsilyl; LiHMDS, lithium hexamethyldisilazide; Cy, cyclohexyl.

pyridine in 65% yield for the two steps. All intermediates were First, we modeled relationships of the epothilones with characterized by satisfactory NMR (1H and 13C) and high- nonataxel. The high temperature molecular dynamics (MD) resolution MS data. study of epothilone B (SYBYL 6.4) revealed several low-energy Characterization Data for Hybrid Construct 6. mp 190– conformers that were evaluated with regards to superimposi- ␣ 22 Ϫ 1 192°C; [ ]D 109° (c 0.11, CHCl3); H NMR (250 MHz, tion with the template conformation of nonataxel. A combi- ␦ CDCl3) 1.14 (s, 3 H), 1.26 (s, 3 H), 1.38 (s, 9 H), 1.55 (m, 1 nation of template forcing (DISCOVER 95.0, Molecular Simu- H), 1.70 (s, 3 H), 1.89 (m, 1 H), 1.91 (s, 3 H), 2.14 (s, 3 H), 2.23 lations), restrained molecular dynamics, and unrestrained en- (s, 3 H), 2.25–2.55 (m, 4 H), 2.77 (m, 1 H), 2.99 (d, J ϭ 8.0 Hz, ergy minimizations (SYBYL 6.4) on these conformations were 1 H), 3.03 (s, 1 H), 3.79 (d, J ϭ 7.1 Hz, 1 H), 4.00 (d, J ϭ 7.7 used to iteratively refine a large number of overlays to obtain Hz, 1 H), 4.07 (m, 1 H), 4.24 (d, J ϭ 8.4 Hz, 1 H), 4.42 (m, 1 the best fit.** Fig. 4b illustrates the excellent topological H), 4.42 (d, J ϭ 8.3 Hz, 1 H), 4.55 (d, J ϭ 9.7 Hz, 1 H), 4.67 homology between epothilone B (2b) and nonataxel (5). The (d, J ϭ 15.0 Hz, 1 H), 4.72 (d, J ϭ 15.0 Hz, 1 H), 4.90 (d, J ϭ C-1 to C-6 segment of 2b corresponds to the ‘‘southern’’ 8.1 Hz, 1 H), 5.69 (m, 3 H), 5.69 (t, J ϭ 6.7 Hz, 1 H), 6.26 (s, hydrophobic surface that interacts at the tubulin-binding site, ϭ ϭ and the C-15 side chain is seen to correspond with the t-Boc 1 H), 6.98 (d, J 8.5 Hz, 1 H), 7.03 (t, J 7.6 Hz, 1 H), 7.47 Ј (t, J ϭ 7.9 Hz, 1 H), 7.68 (d, J ϭ 7.4 Hz, 1 H); 13C NMR (62.9 group at C-3 Nof5. The essential feature of the proposal is the ␦ excellent overlap of (i) the C-2 3-methyl-2-butenoate and C-3Ј MHz, CDCl3) 9.8, 15.3, 20.9, 22.0, 22.9, 26.6, 28.2, 28.4, 29.7, 32.2, 35.5, 37.7, 42.7, 45.7, 51.7, 58.6, 70.4, 71.9, 73.1, 75.5, 75.8, 2-methyl-1-propenyl substituents of 5 with the epothilone 76.4, 77.8, 80.0, 80.9, 84.8, 113.6, 121.3, 128.7, 129.0, 132.1, macrolide core, and (ii) the C13-N-t-Boc moiety with the thiazole ring side chain. 133.0, 133.7, 142.8, 155.6, 168.6, 169.7, 171.4, 203.9. High- The observed structural homology between epothilone B resolution MS (fast atom bombardment) m͞z calculated for ϩ and nonataxel in fact nicely accommodates the vast SAR data C H NO ⅐H : 856.3757; found: 856.3756 (⌬ϭ0.1 ppm). 44 57 16 that have been accumulated for the former. For example, inversion at C-3, reduction at C-5, or removal of functionality RESULTS AND DISCUSSION at any of the positions from C-3 to C-8 in epothilone results in abrogation of both cytotoxicity and tubulin-binding ability Nonataxel (5) is a nonaromatic paclitaxel mimic lacking the (24). The sensitivity of this region to structural alteration conventional N-benzoylphenylisoserine side chain at the accords well with its overlay with the highly sensitive C-3Ј and C-13␣ hydroxyl of baccatin III and the benzoyl ester at the ␣ C-2 alkyl substituents of nonataxel. The ring conformation of C-2 hydroxyl group. Despite its structural difference, nona- epothilone B may potentially be stabilized by hydrogen bond- taxel exhibits 2- to 8-fold higher activity than paclitaxel in ␮ ing between the C-1 carbonyl oxygen and C-3 hydroxyl hydro- various cytotoxicity assays (7) and at 10 M enhances tubulin gen, which could logically explain the loss of activity observed polymerization to the same extent as paclitaxel at that con- on epimerization at the C-3 position (24). Similarly, the centration. Nonataxel, in contrast to paclitaxel, is amenable to deleterious consequence of excision of the C-9 methylene a conformationally defining NMR analysis. High-temperature (producing a 15-membered macrolide) is consistent with the restrained molecular dynamics (RMD) study (SYBYL 6.4, Tri- anticipated change in conformation. The previously demon- pos) on nonataxel, maintaining distance restraints obtained strated need for the thiazole sector is well accommodated by from the two-dimensional NOE spectroscopy experiments in the proposed common pharmacophore, since the aryl sector of ͞ DMSO-d6 D2O (3:1) revealed a very specific relative orien- 2 overlays with the critical C-13 acyl side chain. Also, deletion tation of the C-3Ј and C-2 alkenyl side chains for the major of the olefin spacer, between the aryl and the macrolide conformation of nonataxel. The solution conformation of sectors, results in a predictable loss of activity. The C-12 methyl nonataxel differs only slightly from the previously described group in epothilone B is not defined in a clear way as part of conformation of paclitaxel (20–23) (see Fig. 4a) in terms of a the common pharmacophore. In practice, it can be deleted greater compactness of the hydrophobic clustering of the side (compare epothilone A) or further extended to a propyl group chains. It was expected that the nonaromatic groups in the while retaining good cytotoxicity (24). nonataxel template would allow for better mapping to the ring An important triumph of the proposed common pharma- systems of 2 and 3. cophore is that it accommodates the nonessential nature of the Downloaded by guest on September 27, 2021 Chemistry: Ojima et al. Proc. Natl. Acad. Sci. USA 96 (1999) 4259

FIG.4. (a–d) Overlay of nonataxel (4, cyan) with (a) paclitaxel (1), (b) epothilone B (2b), (c) eleutherobin (3), and (d) discodermolide (4) (all in yellow). Designators A, B, and C correspond to regions of common overlap (also see Fig. 1). (e) Overlay of paclitaxel (1, magenta), nonataxel (4, cyan), epothilone B (2b, yellow), and eleutherobin (3, orange) demonstrating the common pharmacophore (discodermolide omitted for clarity).

C12–C13 epoxide in the epothilones. Thus, this oxido oxygen with our model is somewhat arbitrary. However, we can readily points away from the overlapping structural terrain and is identify a conformation that adheres to our common pharma- therefore not central to the tubulin-binding cytotoxicity phe- cophore. The overlay of this conformation with nonataxel (5) nomena. In fact, the 12,13-desoxy version of epothilone B is (Fig. 4d) shows the efficient overlay in the southern hydro- showing a far more promising profile of in vivo activity than phobic surface traced by the C-11 to C-19 fragment of 4. The either 2b or paclitaxel (25, 26). Similarly, the more severely C-20 to C-24 diene-containing moiety of 4 overlays nicely with distorted 12,13-(E) analogue of 12,13-desoxyepothilone B also the C3Ј-N-t-Boc group of 5. Although not conclusive, this retains excellent tubulin-binding activity. overlay suggests that discodermolide generally conforms to the Next, we turned our attention to the structural homology common pharmacophore proposal. between eleutherobin and nonataxel. As Fig. 4c shows, the Fig. 4e summarizes the overlays of paclitaxel, nonataxel, tricyclic 4,7-oxaeunicellane core of eleutherobin fits neatly into epothilone B, and eleutherobin, which facilitates visualization the cavity between the baccatin core and the southern hydro- of our proposed pharmacophore. One other pharmacophore phobic surface of nonataxel. The largely hydrophobic C-8 to hypothesis for epothilone B and paclitaxel had previously been C-14 segment of eleutherobin traces the southern hydrophobic reported (18). Our pharmacophore proposal uniquely suc- surface, whereas the C-8 N-(6Ј)-methylurocanic acid ester side ceeds in explaining the key structural requirements for the chain closely corresponds to the t-Boc group at C-3ЈNofthe biological activity of paclitaxel—i.e., the C-2 benzoate and the C-13 side chain of nonataxel. Two essential points are imme- two C-13 side-chain elements, (i) the C-3Ј-N-benzoyl that diately recognized: (i) the overlap for the C13-Nt-Boc with the crucially overlaps with the necessary thiazole (or similar C-8 N-(6Ј)-methylurocanic acid ester side chain, and (ii) the aromatic moiety) of epothilone, and (ii) the C-3Ј phenyl group. potentially nonessential role played by the C-3 sugar side chain, In this regard, the previously reported model (18) does not which overlays with the baccatin core rather than the essential explain the critical nature of the C-3Ј phenyl group toward the pharmacophore of nonataxel. tubulin binding and cytotoxicity of paclitaxel, and it places Again, the pharmacophore proposal accommodates the undue importance on the nonessential C-10 acetyl group. At preliminary SAR data available for eleutherobin from these the same time, our proposal also accounts for the activity of laboratories (27). For example, replacement of the L-arabinose sugar-deletion analogues or C-8 side-chain analogues of eleu- moiety with a simple acetyl group or the D-arabinose moiety therobin. Most importantly, our proposal logically suggests the (neo-eleutherobin) results in compounds that retain tubulin linkage between the C-3Ј phenyl group and the C-2␣ benzoyl polymerization ability as well as significant, albeit reduced, group that has allowed us to validate our model through cytotoxicity (27). Nicolaou and co-workers have also recently manufacture of an active hybrid construct described below. reported sarcodictyings—i.e., sugar-free eleutherobin ana- Our model paves the way for designing the third-generation logues—with tubulin polymerization abilities comparable to taxoids that could be essentially baccatin-free, or ‘‘hybrids’’ those of paclitaxel, the epothilones, and elutherobin (28). In integrating the structures of paclitaxel and the new antitumor sharp contrast to the C-3 modifications, the deletion of the C-8 agents. The ring systems of eleutherobin and epothilone B side chain results in abrogation of tubulin polymerization clearly suggested that the nonataxel C-2 and C-3Ј substituents ability and diminution of cytotoxic potency by nearly three can be linked to create a conformational constraint (Fig. 5). On orders of magnitude (27). These results clearly indicate the the basis of the model, we have designed and synthesized critical role of the C-8 side chain, and the expendability of the cytotoxic hybrid constructs containing 16-, 17-, or 18- C-3 sugar moiety, in complete agreement with the common membered macrocycles. pharmacophore. Indeed, by this thinking we were led to synthesizing com- Modeling studies of discodermolide (4) have shown it to be pound 6 (SB-TE-1120). This construct exhibited submicromo- ␮ a very flexible structure. Given the flexibility of the disco- lar level IC50 values (0.39 M against the human breast cancer dermolide system, the selection process required to interface cell line MDA-435͞LCC6-WT) and 37% activity as compared Downloaded by guest on September 27, 2021 4260 Chemistry: Ojima et al. Proc. Natl. Acad. Sci. USA 96 (1999)

FIG. 5. Structural correlation of the hybrid construct SB-TE-1120 (6) with epothilone B (2b) and eleutherobin (3). Designators A, B, and C correspond to regions of common overlap. The overlay of construct 6 (orange) with epothilone B (2b, yellow) and nonataxel (5, cyan) is shown on the right.

with paclitaxel in the tubulin polymerization assay. Fig. 5 8. Shiff, P. B., Fant, J. & Horwitz, S. B. (1979) Nature (London) 277, shows the overlay of construct 6 with nonataxel, demon- 665–667. strating good overlay in the region corresponding to the C-2␣ 9. Jordan, M. A., Toso, R. J. & Wilson, L. (1993) Proc. Natl. Acad. substituent although the overlay near the C-3Ј position, Sci. USA 90, 9552–9556. including the C-3Ј-N-t-Boc group has not reached a desirable 10. Bollag, D. M., McQueney, P. A., Zhu, J., Hensens, O., Koupal, level. Although details of the synthesis and SAR of these L., Liesch, J., Goetz, M., Lazarides, E. & Woods, C. M. (1995) compounds will be published elsewhere, it is relevant to this Cancer Res. 55, 2325–2333. 11. Kowalski, R. J., Giannakakou, P. & Hamel, E. (1997) J. Biol. discussion to note that the cytotoxicity and tubulin-binding Chem. 272, 2534–2541. ability of these compounds show high sensitivity to steric ␣ 12. Lindel, T., Jensen, P. R., Fenical, W., Long, B. H., Casazza, bulk of the substituent at the C-2 position (as observed in A. M., Carboni, J. & Fairchild, C. R. (1997) J. Am. Chem. Soc. the nonaromatic taxoid series (7)), size and flexibility of the 119, 8744–8745. macrocycle, and presence of heteroatom(s) in the lower 13. ter Haar, E., Kowalski, R. J., Hamel, E., Lin, C. M., Longley, surface of the macrocycle (corresponding to the electron R. E., Gunasekera, S. P., Rosenkranz, H. S. & Day, B. W. (1996) density on the lower surface of epothilone B). The prelim- Biochemistry 35, 243–250. inary SAR data for this series of compounds are thus highly 14. Kowalski, R. J., Giannakakou, P., Gunasekera, S. P., Longley, consistent with the common pharmacophore model, thereby R. E., Day, B. W. & Hamel, E. (1997) Mol. Pharmacol. 52, supporting the predictive power of our proposal.‡‡ Further 613–622. improvement in design will facilitate development of the next 15. Giannakakou, P., Sackett, D. L., Kang, Y., Zhan, Z., Buters, generation of tubulin-directed anticancer agents based on J. T. M., Fojo, T. & Poruchynsky, M. S. (1997) J. Biol. Chem. 272, this approach. 17118–17125. 16. Kowalski, R. J., ter Haar, E., Longley, R. E., Gunasekera, S. P., Lin, C. M., Day, B. W. & Hamel, E. (1995) Mol. Biol. Cell 6, ‡‡Ojima, I., Lin, S., Inoue, T., Walsh, J. J., Borella, C. P. & Chakra- varty, S. (1998) 216th American Chemical Society National Meeting, 368a. Boston, August 23–27, 1998; abstr. ORGN 380 and 464, and MEDI 17. Hung, D. T., Chen, J. & Schreiber, S. L. (1996) Chem. Biol. 3, 313. 287–293. 18. Winkler, J. D. & Axelsen, P. H. (1996) Bioorg. Med. Chem. Lett. 6, 2963–2966. This research was supported by grants from the National Institutes 19. Ojima, I., Lin, S., Chakravarty, S., Fenoglio, I., Park, Y. H., Sun, of Health (I.O., S.B.H., and S.J.D.). I.O. and S.C. also acknowledge the C., Appendino, G., Pera, P., Veith, J. M. & Bernacki, R. J. (1998) support from the National Science Foundation for the NMR Facilities J. Org. Chem. 63, 1637–1645. at Stony Brook. 20. Vander Velde, D. G., Georg, G. I., Grunewald, G. L., Gunn, C. W. & Mitscher, L. A. (1993) J. Am. Chem. Soc. 115, 11650– 1. Rowinsky, E. K., Cazenave, L. C. & Donehower, R. C. (1990) 11651. J. Natl. Cancer Inst. 82, 1247–1259. 21. Williams, H. J., Scott, A. I., Dieden, R. A., Swindell, C. S., 2. Suffness, M. (1995) Taxol: Science and Applications (CRC, Boca Chirlian, L. E., Francl, M. M., Heerding, J. M. & Krauss, N. E. Raton, FL). (1994) Can. J. Chem. 252–260. 3. Georg, G. I., Chen, T. T., Ojima, I. & Vyas, D. M. (1995) Taxane Anticancer Agents: Basic Science and Current Status (Am. Chem. 22. Mastropaolo, D., Camerman, A., Luo, Y., Brayer, G. D. & Soc., Washington, DC). Camerman, N. (1995) Proc. Natl. Acad. Sci. USA 92, 6920– 4. Ojima, I., Kuduk, S. D. & Chakravarty, S. (1998) Adv. Med. 6924. Chem. 4, 69–124. 23. Ojima, I., Kuduk, S. D., Chakravarty, S., Ourevitch, M. & Be´gue´, 5. Ojima, I., Slater, J. C., Michaud, E., Kuduk, S. D., Bounaud, J.-P. (1997) J. Am. Chem. Soc. 119, 5519–5527. P.-Y., Vrignaud, P., Bissery, M.-C., Veith, J., Pera, P. & Bernacki, 24. Su, D., Balog, A., Meng, D., Bertinato, P., Danishefsky, S. J., R. J. (1996) J. Med. Chem. 39, 3889–3896. Zheng, Y., Chou, T., He, L. & Horwitz, S. B. (1997) Angew. 6. Ojima, I., Slater, J. S., Kuduk, S. D., Takeuchi, C. S., Gimi, R. H., Chem. Int. Ed. Engl. 36, 2093–2096. Sun, C.-M., Park, Y. H., Pera, P., Veith, J. M. & Bernacki, R. J. 25. Chou, T., Zhang, X., Balog, A., Su, D., Meng, D., Savin, K., (1997) J. Med. Chem. 40, 267–278. Bertino, J. R. & Danishefsky, S. J. (1998) Proc. Natl. Acad. Sci. 7. Ojima, I., Kuduk, S. D., Pera, P., Veith, J. M. & Bernacki, R. J. USA 95, 9642–9647. (1997) J. Med. Chem. 40, 279–285. 26. Chou, T. C., Zhang, X. G., Harris, C. R., Kuduk, S. D., Balog, A., Downloaded by guest on September 27, 2021 Chemistry: Ojima et al. Proc. Natl. Acad. Sci. USA 96 (1999) 4261

Savin, K. & Danishefsky, S. J. (1998) Proc. Natl. Acad. Sci USA Hosokawa, S., Vourloumis, D. & Li, T. (1998) Angew. Chem. Int. 95, 15798–15802. Ed. Engl. 37, 1418–1421. 27. McDaid, H. M., Bhattacharya, S. K., Chen, S., He, L., Shen, H., 29. Nogales, E., Wolf, S. G. & Downing, K. H. (1998) Nature Gutteridge, C. E., Horwitz, S. B. & Danishefsky, S. J. (1999) (London) 391, 199–203. Cancer Chemother. Pharmacol., in press. 30. Ho¨fle, G., Bedorf, N., Steinmetz, H., Schomburg, D., Gerth, K. & 28. Nicolaou, K. C., Kim, S., Pfefferkorn, J., Xu, J., Ohshima, T., Reichenbach, H. (1996) Angew. Chem. Int. Ed. Engl. 35, 1567–1569. Downloaded by guest on September 27, 2021